Thought disorder has frequently been described either as a disturbance of associative response hierarchies or as an inability to deal with objects or events as representative of classes. The tendency of normal subjects to organize free recall by clustering associatively or conceptually related words suggested the recall task as a means to investigate these views of thought disorder. Two groups of psychiatric patients, one with high scores on Lovibond's version of the OST, one with low (thought-disordered and non-thought-disordered, respectively), were presented with randomly ordered word lists containing pairs related associatively and pairs related taxonomically. Thought-disordered and non-thought-disordered groups were not differentiated by any tendency to cluster more of one kind of pair than another. Rather, the thought-disordered showed a general lack of organization in recall, and significantly less improvement over trials in amount recalled. The results do not support the view that thought disorder is specifically a disorder of concept use or conceptual mediation. Research needs to be addressed to variables which affect the activation or inhibition of mediational links in thinking.